.jonas
Things connecting people
Archive for Network analysis
March 15, 2008 at 11:27 am · Filed under Network analysis
My friend Timme and I had some interesting discussions about his article (in progress) for the Journal of Computer Mediated Communication. He chose not to include my input about link typologies and their connection to conversion of social capital between offline and online arenas (and vice versa), but I think its worth publishing an early draft here for critique: The article is a study of linking in the Kforum business network for communication and marketing professionals in Denmark. On the Kforum network users categorize their relations to others in four types of relations. From the perspective of an ego-user the four types of links are:
- Inbound - unconfirmed: Link is established by other user with no confirmation by ego.
- Inbound - confirmed: Link established by other user confirmed by ego.
- Outbound - unconfirmed: Link established by ego with no confirmation by other user.
- Outbound - confirmed: Link established by ego confirmed by other user.
These four types of social link relations are visible to all other users on each profile. This means that both social inclusion and exclusion is visible to all other users in the network whenever and wherever a user chooses to navigate unconditioned by mutual confirmation as on Facebook and linkedIn. However, even though more visible on Kforum, these four basic link types are prevalent in most online networks.
As illustrated in the table, the different prevalent link typologies represent different mods of conversion of social capital between online and off-line arenas. People with mostly inbound- links are typically either persuing a defensive strategy passively protecting their offline social capital by simply not linking (inbound - unconfirmed) or attemting to construct online social capital to be converted to offline arenas by contructing new relations through confirmed links (inbound - confirmed). The resulting strategy, or behavioral modes, for networkers with mostly inbound links are passive or constructive.
Networkers with mostly outbound links are typically either actively linking to people they know already from offline arenas, converting their accumulated social to the online network (Outbound - confirmed) or spamming the network hoping to create new relations at random thus acummulating new social capital through ultra-weak ties (Outbound - unconformed). [Inconsistency in the table to be rectified shortly. Sorry..] The resulting bahaviours are here called converting and signaling.
So, what kind of networker are you? Passive, constructive, converting or signaling?
November 19, 2007 at 1:05 pm · Filed under Network analysis, New media, Systems Theory
After an irresistible surge of peer pressure I finally gave in and created a facebook profile this weekend. My initial resentment towards putting too much energy into maintaining yet another web 2.0 profile was spawned by the seeming lack of a shared theme or, in the parlour of computer science, a shared object like music on mySpace.com, communication on kommunikationsforum.dk or growing up rural on bondeknold.com. Normally I would expect every self-organizing web community to be exclusively differentiated and organized around a relatively narrow thematic code, but facebook doesn’t seem to work that way. Or does it?
Where other 2.0 apps organize around differentated themes, they don’t show off the users’ social capital to users that are not a part of your immidiate network. An intriguing paper in the Journal of Computer Mediated Communication by Nicole Ellison et al. show that there is a connection between facebook use and social capital among students at the Michigan State University. Ellison’s study reveals a crucial clue to understanding what facebook does for me: it shows off my social capital (friendly connections) to a wide audience.
Facebook is to social capital what bling is to money. In stead of showing off your riches, it shows off how well connected you are. They even made little accessory apps to put on your profile to add a little extra ice. My favorite is called the ‘Friend Wheel‘ and it shows acircular ego network model as proposed by Duncan Watts and Steven Strogatz in Watts’ now infamous book ‘Six Degrees - The Science of a Connected Age‘. Here is my current friend wheel bling:
Blue: People I grew up with
Turquoise: High School friends
Green University alumni
Yellow: Work colleagues
Red: Miscellaneous aquantances
Your friends are iced out bling and chrome rims on a big ass SUV!
November 15, 2007 at 8:53 am · Filed under Complexity theory, Network analysis
When browsing the litterature on Social Network Analysis (SNA) most actual analyses contain only very few nodes, like 20-100 at most. While these analyses inform the methodology of SNA they do not tell us much about complex networks. 100 nodes can only reproduce so much complexity, so my friend and colleague Timme and I want to do better. We want to do a SNA of a network of 2752 nodes in the K-profil network. Now that’s complexity. But also it proves a huge challenge.
Lesson #1: Complexity is really complex
November 1, 2007 at 3:18 am · Filed under Network analysis
I just stumbled across this snappy poster by the boys at FAS.research in Austria (Harald Katzmair & Helmut Neundlinger) showing the 2006 world cup match between Germany and Sweeden in terms of Social Network Analysis. The connections represent the passes, and, surprisingly enough, it gives a great overview of which players delivered a good game and who were invisible on the court. Judge for yourself:
Actually it should be relatively simple: a football team network has relatively low degree of complexity, and the analysis is pretty straight forward once you have the passing statistics. What could be really interesting to the coaches is to make a network analysis of each team including goals as a node, lost balls a red aura etc. That way the analysis could relatively easily be applied as a decision-making instrument on the sidelines.
I wonder if there are any other unlikely domains out there just waiting to be modelled.. Any suggestions?
Network Analysis everywhere!
